Data and code for "Linked biases in terrestrial water storage trend magnitude, variability, and drought in global water models"

Le résumé fourni par la source

GWM-TWS-GRACE v1.0.1 Data, analysis code, and figure source files for the manuscript Global water models underestimate the magnitude of terrestrial water storage change and associated drought characteristics. This peer-review release evaluates terrestrial water storage (TWS) simulated by 10 global water models against three GRACE/GRACE-FO RL06 mascon products in 180 river basins from April 2002 through December 2019 (213 months). Scientific scope The package evaluates: TWS trend direction and trend-magnitude bias; deseasonalized TWS variability (TWSV); storage-deficit events standardized by the GRACE calendar-month SD; joint underestimation across trend, variability, and drought metrics; differences associated with climate aridity, human activity intensity (HAI), and nine model process-representation classes. The 10 models are CWatM, H08, HydroPy, WaterGAP2-2e, WEB-DHM-SG, JULES-ES, JULES-W2, MIROC-INTEG-LAND, LPJmL5-7-10-fire, and ORCHIDEE-MICT. The complete archive contains 32 available model-forcing-society configurations. Figures 5-7 use the balanced set of 10 GSWP3-W5E5/histsoc configurations, one per model; they do not use all 32 configurations. Canonical definitions Trend: OLS slope of the original monthly TWS series, in km3 yr-1. Trend-magnitude bias: |beta_model| - |beta_GRACE|. Relative trend-magnitude bias: 100 * (|beta_model| - |beta_GRACE|) / |beta_GRACE|. TWSV: sample SD of the deseasonalized monthly TWS anomaly, in km3. Drought index: both model and GRACE anomalies divided by the GRACE calendar-month SD. Drought event: a maximal contiguous run with Z < -1. Long deficit: a drought event lasting at least 12 months. Recovery time: months from the end of a GRACE event to the first subsequent month with Z > -0.5. Event match: model and GRACE event intervals overlap by at least one month. See docs/methodology.md for the complete analysis sequence and interpretation limits. Repository layout gwm-tws-grace-v1.0.1/ |-- README.md |-- CITATION.cff |-- LICENSE, LICENSE-CODE, LICENSE-DATA |-- environment.yml, requirements.txt |-- reproduce.py, reproduce.sh |-- data/ | |-- basin_polygons/ | |-- grace_processed/ | |-- model_bias/ | |-- figures/ reference main-figure JPG files | `-- supplementary/ reference supplementary figures |-- scripts/ | |-- common/ | |-- figure1/ ... figure7/ | `-- supplementary/ |-- docs/ `-- tests/ Installation The recommended environment is: conda env create -f environment.yml conda activate gwm-tws-grace The pip alternative is: python -m venv .venv python -m pip install -r requirements.txt Validation and reproduction Run structural and scientific consistency checks: python reproduce.py --check Recalculate and plot the non-map analyses: python reproduce.py --figures 4 5 6 7 Reproduce all main and supplementary figures: python reproduce.py --all --supplementary Generated products are written to outputs/; source data under data/ are not modified. reproduce.sh is a small POSIX wrapper around reproduce.py. Data and code boundaries This archive contains processed basin-scale data and figure source tables, not the complete raw GRACE or ISIMIP3a archives. Each model supplies a total TWS output variable; model TWS was not reconstructed by summing storage components. The process matrix describes model structural capabilities and is not a causal attribution of individual modules. The variable-level inventory is docs/data_dictionary.csv. Data provenance and redistribution qualifications are described in data/README.md. Citation Use the citation in CITATION.cff. The manuscript journal and article DOI should be added only after they are assigned. Licenses Code under scripts/, tests/, and reproduce.py is licensed under the MIT License. Package-authored processed tables, documentation, and figures are licensed under CC BY 4.0, subject to the original terms of the source datasets. See LICENSE, LICENSE-CODE, and LICENSE-DATA. Contact Corresponding author: Wei Wei, Northwest Normal University, weiwei@nwnu.edu.cn.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.